CORRECTION OF LOGISTIC REGRESSION RELATIVE RISK ESTIMATES AND CONFIDENCE INTERVALS FOR MEASUREMENT ERROR: THE CASE OF MULTIPLE COVARIATES MEASURED WITH ERROR

摘要: If several risk factors for disease are considered in the same multiple logistic regression model, and some of these measured with error, point interval estimates relative corresponding to any may be biased either toward or away from null value. A method is provided correcting obtained measurement error one more continuous variables. The requires a separate validation study estimate coefficients multivariate linear model relating surrogate variables vector true factors. Similar methods have been suggested by other authors, but none provides means confidence intervals which include component variability due estimation parameters study. An example prospective dietary fat, calories, alcohol relation breast cancer, questionnaire used assess nutrients. Before age-adjusted 25 g increment intake was 1.33 (95% (CI) 1.14-1.55); after increased 1.62 CI 1.23-2.12). Similarly, 10 saturated fat intake, 0.94 0.83-1.06); 0.84 0.59-1.20). These results indicate that failure find substantial positive association between cancer cannot explained alcohol.